1. Field of the Invention
The present invention relates to a semiconductor device and a method of fabricating the same, and more particularly to an improvement of gate electrodes of an n-type MIS transistor and a p-type MIS transistor.
2. Description of the Related Art
Miniaturization of devices is indispensable in enhancing the performance of MIS transistors. However, a silicon oxide film, which is currently used as a gate insulation film, has a low dielectric constant, and thus the capacitance of the gate insulation film cannot be increased. In addition, since a polysilicon used as a gate electrode has a high resistivity, it is difficult to decrease the resistance of the gate electrode. To solve these problems, there is an idea that a high dielectric constant material is used for the gate insulation film and a metallic material is used for the gate electrode.
However, these materials have drawbacks in that the heat resistance thereof is lower than that of currently used materials. A damascene gate technique has been proposed as a technique wherein a gate insulation film and a gate electrode can be formed after a high-temperature process is carried out.
In a case where a metal is buried as gate electrodes by the damascene gate technique, the gate electrodes of an n-type MISFET and a p-type MISFET are formed of a single metal and the work function of the gate electrodes is fixed. Thus, unlike the case of polysilicon gates, it is not possible to optimize threshold values by forming different gate electrodes in n-type and p-type devices. A dual metal gate process is thus required in order to form gate electrodes of different materials in n-type and p-type devices.
The inventors previously filed a patent application (application Ser. No. 09/559,356) for a technique for forming different metal gate electrodes in n-type and p-type devices. The steps of a process of fabricating a semiconductor device according to the method of this application will now be described with reference to FIGS. 3A to 3J.
To start with, a device isolation region 101 is formed on a silicon substrate 100 by means of an STI (shallow trench isolation) technique, etc. An p-well 102 is formed in a formation region of an n-type MISFET and a n-well 103 is formed in a formation region of a n-type MISFET. A dummy gate lamination structure is then formed as a dummy gate that is to be removed later, by means of techniques of oxidation, CVD, lithography, RIE, etc. The dummy gate lamination structure comprises a gate oxide film 104, which is, e.g. about 6 nm thick, a polysilicon 105, which is about 150 nm thick, and a silicon nitride film 106, which is about 50 nm thick. An extension diffusion layer region 107 is formed using an ion implantation technique. A gate side wall 108 with a thickness of about 40 nm, which is formed of a silicon nitride film, is formed by CVD and RIE techniques.
In FIG. 3B, a source/drain diffusion layer 109 is formed by an ion implantation technique. Then, using the dummy gate as a mask, a silicide 110 of cobalt, titanium, etc. with a thickness of about 40 nm is formed only in the source/drain region by means of a salicide process technique.
In FIG. 3C, a silicon oxide film, for example, is deposited by CVD as an interlayer film 111. The silicon oxide film is then flattened by CMP to expose surfaces of the silicon nitride film 106 and gate side wall 108 at an upper part of the dummy gate.
In FIG. 3D, the silicon nitride film 106 at the upper part of the dummy gate is selectively removed relative to the interlayer film 111 by using a phosphoric acid, for instance. At this time, the gate side wall 108 at the side wall of the gate is also etched away to a level equal to the level of the polysilicon 105. Subsequently, the polysilicon of the dummy gate is selectively removed relative to the interlayer film 111 and the gate side wall 108 of the silicon nitride film by means of, e.g. a radial atom etching technique. Thus, a gate trench 112 is created. The dummy gate oxide film 104 is provided at the bottom of the gate trench 112.
In FIG. 3E, the dummy gate oxide film 104 is removed by a wet process using hydrofluoric acid, etc., thereby exposing the p-well 102 or n-well 103 at the bottom of the gate trench 112.
A gate insulation film 113 of, e.g. a hafnium oxide film is formed as a high dielectric constant insulator over the entire surface of the resultant structure.
In FIG. 3F, a hafnium nitride film 114, as an example of metal having a work function of 4.6 eV or less, is formed by CVD or sputtering with a thickness of about 10 nm, or preferably less than 10 nm, on the entire surface of the resultant.
The steps of FIGS. 3A to 3F are carried out for both the n-type MIS transistor formation region and p-type MIS transistor formation region, but these Figures show only one of these regions. As regards the subsequent steps, FIGS. 3G to 3J show both of the n-type MIS transistor (n-type MISFET) and p-type MIS transistor (p-type MISFET).
In FIG. 3G, that portion of a resist 115, which lies in the p-type MISFET region, is removed by lithography.
In FIG. 3H, wet etching is performed using hydrogen peroxide solution, thereby removing the hafnium nitride film 114 from the p-type region alone. At this time, the gate insulation film 113, which is the hafnium oxide film, is not etched since it is insoluble in the hydrogen peroxide solution.
In FIG. 3I, the resist 115 is removed, and tantalum nitride 116, as an example of a material having a work function of 4.6 eV or more, is deposited with a thickness of at least about 10 nm.
In FIG. 3J, aluminum 117 is deposited as a low-resistance gate electrode material on the entire surface of the resultant by means of sputtering or CVD. Then, the aluminum is subjected to CMP, thus burying the aluminum 117 in the gate trenches.
A CMISFET is fabricated through the above-described steps, which has gate electrode structures comprising, respectively, an n-type lamination structure of the hafnium nitride film 114, tantalum nitride 116 and aluminum 117, and a p-type lamination structure of the tantalum nitride film 116 and aluminum 117. Accordingly, the threshold values can be optimized since the work function of the gate electrode of the n-type device is 4.6 eV or less and the work function of the gate electrode of the p-type device is 4.6 eV or more.
This structure has a problem, however. FIGS. 4A and 4B are enlarged views of the gate electrode portions of the n-type MISFET and p-type MISFET. In the n-type MISFET, a width LA1 of the aluminum of the gate electrode is expressed byLA1=LG−2×LTaN−2×LHfNwhere LA1 is a width of the aluminum 117, LG is a gate length, LTaN is a width of the tantalum nitride film 116, and LHfN is a width of the hafnium nitride film 114.
The tantalum nitride film 116 functions to control the work function of the gate electrode, and also serves as a barrier metal for preventing the upper electrode, i.e. the aluminum 117, from diffusing in the gate insulation film. Accordingly, in view of the gate breakdown voltage and reliability, it is necessary that the thickness of the tantalum nitride film 116 be at least about 10 nm or more.
However, in the case where the gate length (LG) is 40 nm or less, if the film thickness (LTaN) of the tantalum nitride film 116 is 10 nm and the thickness (LHfN) of the hafnium nitride film 114 is 10 nm, the width (LA1) of the aluminum 117 would be 0 nm. Thus, if the gate length (LG) is 40 nm or less, it is impossible to bury the aluminum 117. As a result, the gate resistance greatly increases and a high-performance CMISFET cannot be fabricated. The thickness of the hafnium nitride can be reduced to about 1 nm since the work function alone needs to be controlled. However, in this case, too, the width (LA1) of the aluminum 117 would become 0 nm, when the gate length is 20 nm.
As has been described above, in the semiconductor device with the damascene gate structure, if materials having different work functions are used in the n-type MISFET and p-type MISFET, it is not possible to bury an electrode material with low resistance, and a high-performance CMISFET cannot be fabricated.